This is, of course, very opinion based, but I will defer to the general consensus. I am incredibly skeptical of the whole e/acc brand and philosophy, but I'm willing to give them a chance. This seems like they will ship and/or demo their product, but if it's a false alarm and nothing happens, it sucks by default.
I will be voting yes on this, but I promise that I will be resolving this fairly
Update 2025-10-29 (PST) (AI summary of creator comment): The creator has extended the resolution date to allow time to evaluate Extropic's actual chip and simulation library. Resolution will be based on:
Reading the paper
Looking for evaluations of performance
Assessment beyond just the presentation itself
1,000People are also trading
I haven't seen any actual evidence of anything actually running on a thermodynamic chip, either from extropic themselves or from anyone who has posted a photo of the boxes. Am I just not paying attention?
@Bair sorry, I used lazy terminology; let's rephrase as an application running on a conventional CPU but incorporating sampling from the thermal chip to accelerate some aspect of the workload (with benchmark to compare vs. cpu-only). Basically a demonstration of any kind of utility. Perhaps the definition of what would constitute an enlightening demo requires more knowledge of the thermal computing aspects.
@Tomoffer I haven't seen them elaborate on potential use in such hybrid scenarios; the main idea seems to be to scale the complexity of the distributions baked in their chips until it reaches some threshold where direct sampling with little further processing is actually useful.
